Sources
When an AI answer cites a page that isn't yours and isn't a competitor's, it's citing a third party — a review site, a community thread, a category roundup. Sources rolls all of those up by domain, so you can see exactly which outside sites are earning the visibility in your category and whether your brand shows up on them.
What this screen shows
The Sources screen (/[org]/projects/[project]/visibility/sources) is the earned-visibility view: third-party domains AI cites on your prompts — the review sites, communities and roundups earning your category’s visibility. Where Citations lists every cited URL split into yours and competitors’, Sources takes the everything-else — every cited page that belongs to neither you nor a tracked competitor — and groups it by domain. One row per site, all-time, across every prompt and platform.
This matters because a huge share of AI answers lean on exactly these pages. If ChatGPT keeps citing a G2 category listing or a Reddit thread when it answers your prompts, that page is doing the work of shaping the answer — and the fastest way to change the answer is often to change that page, not one of your own. Every source here is paired with a playbook for going to earn or defend that citation.
Third-party sources
18
Earned citations
63
Cited in latest
11
You're absent from
7
(Illustrative values.) Third-party sources counts the distinct outside domains cited on your prompts. Earned citations sums how many times any of them was cited. Cited in latest is how many of those sources appeared in your most recent answers — i.e. still current, not just historical. You’re absent from counts the sources where OptimizeCamp checked the cited page and confirmed your brand isn’t on it — a count of known absences, not of every site (an unchecked source doesn’t count here).
The sources table
Each row is one third-party domain, with five columns:
| Column | What it shows |
|---|---|
| Source | The domain (e.g. g2.com) with a source-type badge, and beneath it the single most-cited URL on that domain — the actual page AI keeps linking, clickable to open it. |
| Cited on | Prompt chips for the tracked prompts this domain was cited on (up to two, with a “+N more”); click one to open that prompt’s answer. |
| Cites | How many separate answers have cited this domain, shown like 12×. |
| You | A presence pill — Present, Absent, or — — for whether your brand is named on the cited page, plus a “vs {competitors}” line — a truncated, hover-for-full list of the competitor brands named on that same cited page — when rivals appear there. |
| Playbook | The play attached to this source — its status and a View playbook / View result link, or a Create playbook button if none exists yet. A Dismiss link hides sources you don’t care about. |
Source types
Each domain is tagged with a source type — how AI tends to use it — classified from a built-in map of well-known sites. The type is what decides which kind of playbook you’ll get, so it’s worth knowing the five:
| Type | What it is | Typical sites |
|---|---|---|
| community | Forums and Q&A where people discuss your category. | Reddit, Quora, Stack Exchange, Hacker News |
| review | Software and product review platforms. | G2, Capterra, TrustRadius, Trustpilot |
| directory | Listings and profiles you can claim. | Product Hunt, Crunchbase, Clutch, Gartner |
| reference | Encyclopedic references. | Wikipedia, Wikidata |
| editorial | The default for anything else — blogs, listicles, news, roundups. | Independent blogs, “best tools” articles, press |
Are you on the page? The presence pill
The You column answers a simple, high-value question: when AI cites this page, does it find your brand on it? OptimizeCamp fetches the most-cited page and checks for your brand name and aliases, giving one of three states:
- Present — your brand was found on the cited page. The citation is already working in your favor; the play here is usually to defend it.
- Absent — the page was fetched and your brand is not on it. This is the gap: AI is citing a page you could be on and aren’t.
- — — unknown. The page hasn’t been checked yet, or the fetch was blocked, timed out, or wasn’t HTML. The pill’s tooltip reads “Not checked yet — use Check presence.”
Check presence on demand
When presence is unknown, the playbook drawer for that source offers a Check presence button that fetches the page live and fills the pill in on the spot — you never have to wait for the background pass to get around to it.
Filtering by what you’re doing about it
A row of filter chips — All, Suggested, In progress, Won — narrows the table by the state of each source’s attached play, so you can jump straight to the sources you haven’t started on, the ones a teammate is working, or the wins. Each chip carries its own count. The play lifecycle behind these states is covered in Earned-media playbooks.
How the list stays current
You never refresh this screen by hand. Sources are rolled up from your existing citation data, and the roll-up happens in the background when you open the page — so a project that predates this feature backfills on its first visit without making you wait. Two light guards keep it efficient:
- The roll-up only recomputes when tracking has produced new data and the last recompute is more than a few hours old.
- Enrichment (source type + presence) fills in a small batch of the top sources per visit, and re-checks presence when it’s more than a week stale. Source type is always set instantly; presence degrades quietly to — if a page can’t be fetched.
Dismissing a source, and what sticks
A source has a simple two-state life: active (shown) and dismissed (hidden after you click Dismiss). Because the roll-up only ever overwrites the count and co-occurrence columns, your dismissal — and any enrichment — is preserved across every recompute: a dismissed source stays dismissed, and a source is never silently dropped from the data.
Note
Sources is a core, always-free screen — no plan gating and no credits to read it, roll it up, enrich it, or check presence. (The one metered thing in this area is AI-drafted outreach copy inside a playbook — see Plans, quotas & credits.)
Before you have any sources
If no third-party domains have been cited on your prompts yet, the screen shows “No third-party sources yet” and points you back to Visibility to run a tracking cycle — sources can only appear once there are answers (and their citations) to roll up.
Key takeaways
- Sources rolls up every third-party domain AI cites on your prompts — everything that isn't yours and isn't a tracked competitor — into one row per site.
- Four tiles summarize it: third-party sources, earned citations, how many were cited in your latest answers, and how many you're confirmed absent from.
- Each domain gets a source type (community, review, directory, reference, editorial) that determines the playbook, and a presence pill (Present / Absent / —) for whether your brand is on the cited page.
- Every source pairs with a playbook to go earn or defend the citation; filter chips (Suggested / In progress / Won) track where each one stands.
- It's a free, always-on screen that rolls up and enriches itself in the background — dismissals and enrichment survive every recompute.
Next step
A source tells you where AI is looking; a play tells you what to do about it. Earned-media playbooks covers the typed off-site plays — claim, review, pitch, defend — and the verification loop that measures whether the work paid off.